SMPS Voltage Drift Problem (TL431)

[azadfalah]

Hello friends
I used the following circuit to make a 48V charger

The problem of this SMPS is that it has different voltage at different temperatures, the changes are about one volt
I understand that the PCB file helps to answer the question, but it is not possible to provide the PCB

How can I make the voltage of this charger stay exactly at 48 volts and not be in the range of 47.5 to 48.5?

I used the following circuit, but with accurate recalculations, I set the output voltage to 48 volts 0.5A with (Current Sense Resistor is 0.1R)

 

[KlausST]

Hi,

In the analog world there is no "exactly 48V". You always have to live with tolerances and noise.

We don't know how much error you can tolerate.
Errors are: absolute accuracy error, drift vs time, drift vs temperature, drift vs load current, noise..

Klaus

 

[andre_luis]

You could consider make experiments with the LM4041, which if I'm not wrong is pin compatible with the TL431, and have a significantly smaller tmperature drift.

 

[D.A.(Tony)Stewart]

Hello friends
I used the following circuit to make a 48V charger

The problem of this SMPS is that it has different voltage at different temperatures, the changes are about one volt
I understand that the PCB file helps to answer the question, but it is not possible to provide the PCB

How can I make the voltage of this charger stay exactly at 48 volts and not be in the range of 47.5 to 48.5?


View attachment 185876

I used the following circuit, but with accurate recalculations, I set the output voltage to 48 volts 0.5A with (Current Sense Resistor is 0.1R)

Where is your real schematic?
First analyze why the thermal stability is poor, then fix it.

There are a few requirements to optimize thermal stability;

1: insulate or isolate the TL431 case temperature from local heating.
2. prevent self heating by limiting power dissipation using Ika = 10 mA with the series R. and achieve optimum Vref. in the middle of your operating temperature range.
3. Check the bandgap Vref is near optimum. 2.495 V
4. Define your error specs and conditions for tolerance and stability and choose a suitable design for Vref.

The TL431 is capable of <= 1% error over all parts and entire temperature range which is a reasonable design spec. However, you stated "How can I make the voltage of this charger stay exactly at 48 volts and not be in the range of 47.5 to 48.5?"

This is a 1 % tolerance, so for a better tolerance, you need to stabilize the temperature by operating with a regulated fixed temperature with thermal feedback or operate at a fix temperature in benign conditions. You could also attempt a thermistor compensated error correction , which may increase the cost of the design with trimmers.

I assume you would have changed the = 820 ohm value for 48 V out, otherwise it would contradict the conditions 2 & 3 above.

 

[BigBoss]

Secondary side resistances should be metal film type with low temperature drift feature.
Voltage and Current adjustment trimmers must also be low temperature drift type and Muti Turn is preferable.
But the best option is to replace these trimmers with fixed resistors because trimmers have general speaking have poor temperature characteristics due to their nature.

 

[Easy peasy]

TL431 don't work too well with a 48V rail - read the data sheet for max supply volts to the TL431

try a 4v7 and a 4148 in series with 5mA through,

this will give you a pretty stable 5.2V ref, that can be powered from the 48V output line

you will need to adjust the Vout sense resistors to match to the new higher Vref

 

[D.A.(Tony)Stewart]

TL431 don't work too well with a 48V rail - read the data sheet for max supply volts to the TL431

It works fine using Vka = Vref as is and just attenuate the feedback more to required ratio.

 

[crutschow]

it has different voltage at different temperatures

What is the temperature range?

 

[D.A.(Tony)Stewart]

Where is your real schematic?
First analyze why the thermal stability is poor, then fix it.

There are a few requirements to optimize thermal stability;

1: insulate or isolate the TL431 case temperature from local heating.
2. prevent self heating by limiting power dissipation using Ika = 10 mA with the series R. and achieve optimum Vref. in the middle of your operating temperature range.
3. Check the bandgap Vref is near optimum. 2.495 V
4. Define your error specs and conditions for tolerance and stability and choose a suitable design for Vref.

The TL431 is capable of <= 1% error over all parts and entire temperature range which is a reasonable design spec. However, you stated "How can I make the voltage of this charger stay exactly at 48 volts and not be in the range of 47.5 to 48.5?"

This is a 1 % tolerance, so for a better tolerance, you need to stabilize the temperature by operating with a regulated fixed temperature with thermal feedback or operate at a fix temperature in benign conditions. You could also attempt a thermistor compensated error correction , which may increase the cost of the design with trimmers.

View attachment 185903

View attachment 185904

I assume you would have changed the = 820 ohm value for 48 V out, otherwise it would contradict the conditions 2 & 3 above.

Keep in mind if you draw 10 mA from 48V that the resistor will be 100'C above room temp if you use a 1/2W resistor so choose a 1W R.

That may certainly cause thermal issues near the TL431, so keep well apart.

Also I just noticed, your TL431 will be unstable with 0.1 uF across it when Vak-Vref so remove that and change to 1 nF.

Read the entire datasheet next time pls.

 

[dick_freebird]

Did you notice the curvature of the reference voltage?
It's typically in the 1% range for "bowing", less with thin
film resistors in the bandgap core, more with Pbase due
to the second-order tempco of the resistor body.

48 +/- 0.5V is a 1% supply. The problem, I expect, is in the
expectations.

 

[D.A.(Tony)Stewart]

Did you notice the curvature of the reference voltage?
It's typically in the 1% range for "bowing", less with thin
film resistors in the bandgap core, more with Pbase due
to the second-order tempco of the resistor body.

48 +/- 0.5V is a 1% supply. The problem, I expect, is in the
expectations.

I agree. I suppose if the TL431 was insulated and internally heated with thermistor feedback and heat controlled at 70'C for the 0 to 70'C range, the tempco error could be eliminated and the initial tolerance error nulled out with the feedback pot.